DYNAMIC UPLIFT OF CONCRETE LININGS: THEORY AND CASE STUDIES. E.F.R. Bollaert 1 ABSTRACT

DYNAMIC UPLIFT OF CONCRETE LININGS: THEORY AND CASE STUDIES E.F.R. Bollaert 1 ABSTRACT Dynamic uplift of concrete linings due to severe pressure fluctuations is of major concern to design engineers. The phenomenon has been extensively studied because of major failures in the sixties and seventies. Nevertheless, despite major advances in measurement technology and data acquisition, a safe and economic design method for any kind of concrete lined stilling basin is still missing today. Especially the dynamic or even transient character of pressure pulsations and their two-dimensional spatial distribution above and underneath the lining are not fully assessed and implemented. The present paper briefly outlines the main historic steps of the state-of-the-art in lining design and then presents a new design method for concrete lining uplift. This new physically-based method combines laboratory measurements of net uplift forces, prototype-scaled transient pressure recordings inside artificially generated lining fissures and numerical modeling of air-water pressure pulsations. This defines the time-dependent pressure field underneath the lining. At the upper face of the lining, a spatially distributed but time-averaged pressure field is considered. This finally provides a time-dependent net uplift pressure and impulsion on the concrete lining. Based on the differential equation of a spring-mass system accounting for lining inertia and anchor bar elastic properties, the most critical net uplift impulsion on the lining is then transformed into an equivalent most critical stress of the lining anchor bars, allowing sound dimensioning of the latter. In the following, the main steps of the new design method are presented and illustrated with real-life studies in the western US and in central Switzerland. INTRODUCTION Dynamic uplift of concrete linings due to severe pressure fluctuations is of major concern to design engineers. The phenomenon has been extensively studied because of failures in the sixties and seventies. Nevertheless, despite major advances in measurement technology and data acquisition, a safe and economic design method for any kind of concrete lined stilling basins is still missing today. Especially the dynamic or even transient character of pressure pulsations as a function of their two-dimensional spatial distribution above and underneath the lining is not fully assessed and implemented in existing design methods. Concrete slabs are used as bottom protection linings of spillway stilling basins. Their design focuses on stability and resistance to severe hydrodynamic loadings during floods. The way these extreme loadings are defined, however, has been subject to significant 1 President, AquaVision Engineering Ltd., Chemin des Champs-Courbes 1, CH-1024 Ecublens, SWITZERLAND, admin@aquavision-eng.ch

debates since concrete slabs have at first been used. Initial design rules concentrated on resistance to impact pressures at the slab surface and on sound drainage of static pressure underneath the slabs. The shortcomings of such a design have been experienced during the 1960 s by major damage of several concrete linings. Well-known examples are Malpaso Dam (Mexico) and Karnafuli Dam (Bangladesh). The damage was found to be generated by sudden uplift or detachment of the slabs from the bottom (Bowers and Tsai, 1969; Sanchez and Viscaino, 1973). This uplift occurred at discharges much lower than the design discharge and, as an example, at Karnafuli Dam it was found to be generated by severe pressure fluctuations that may enter the outlets of the drain system and the joints between the slabs. This has stimulated researchers to investigate the presence of dynamic pressures underneath the slabs. These underpressures travel through the joints as pressure waves with celerities that are considered much higher than the travel velocity of the surface pressures from which they originate (Fiorotto and Rinaldo, 1992). They may generate instantaneous net uplift pressures that are able to destabilize the slabs. Determination of underpressures depends on the location and dimensions of both the slab joints and the pressure release pipes of the drainage system. Both types of discontinuities are present in a standard lining design and represent potential entries for underpressures. Therefore, slab joints are most often equipped with water stops that are designed to prevent pressure pulses at the slab surface from entering the joints. These water stops, however, age with time and subsequent flooding. Especially slab vibrations during floods may enhance deterioration of the water stops. Therefore, the use of water stops alone may not be considered as a sufficient countermeasure against slab uplift. Second, depending on the location of the pipe entries and the configuration of the network, drainage pipes may stimulate underpressures. Their influence on the generation and transfer of dynamic pressures underneath slabs is not fully understood yet. Theoretical and experimental research has been performed in the 1980 s and 1990 s to account for instantaneous dynamic underpressures in lining joints (Fiorotto and Rinaldo, 1992; Bellin and Fiorotto, 1995; Fiorotto and Salandin, 2000), however without accounting for transient pressure wave effects. Bellin and Fiorotto (1995) directly measured uplift forces on laboratory scaled concrete slabs of different dimensions and subjected to hydraulic jump impact. The scale of the model did not allow detecting transient waves, however. Other small-scale experiments of uplift pressures on concrete slabs and/or blocks have been performed by Yuditskii (1963) for a ski-jump spillway, by Reinius (1986) for water flowing parallel to the foundation, and by Liu et al. (1998) for the Three-Gorges spillway. Lastly, Melo et al. (2006) proposed a concrete slab design method that is solely based on time-averaged net dynamic uplift pressures. It has to be outlined that all of these tests have been performed at a small scale that does not allow detecting pressure wave phenomena. Recently, prototype-scaled experiments performed in the field of fracturing of rock joints due to high-velocity jet impact (Bollaert

and Schleiss, 2005; Bollaert, 2004) have shown that pressure waves in joints may travel at very low wave celerities, typically 50-200 m/s, due to the presence of air in the water. Hence, transient effects such as wave oscillations and resonance may be relevant when defining extreme pressures underneath concrete linings (Bollaert and Schleiss, 2005). As such, the present paper outlines a new design method for concrete linings of plunge pool stilling basins and illustrates the main steps of the method based on real-life case studies in the western US and in central Switzerland. THEORETICAL BASES OF CONCRETE LINING DESIGN General Design of concrete slabs for hydrodynamic loading focuses on the determination of the maximum possible net uplift pressure (force) and related impulsion. The net uplift pressure is defined by the net difference between surface pressures and underpressures at any given time instant. The net uplift impulsion is determined by the integration over time of the net uplift pressure. In the following, the methodology is outlined using pressures rather than forces (for a unitary slab surface). Pressures occurring at a lining surface can be described by dynamic pressure coefficients. These define the time-averaged pressure field and its spatial distribution over the surface of the lining. Underpressures are defined based on surface pressures that enter the joints between the slabs of the lining and the joints between the lining and its foundation. Slab uplift occurs when the time-averaged or instantaneous pressure (force) differences over and under the slabs are able to generate sufficient impulsion to displace the slab. First, both time-averaged and instantaneous spatial pressure distributions have to be assessed at the surface of the slab. The instantaneous spatial pressure distribution can be estimated by performing large-scale laboratory measurements, which define the spatial correlation of the pressure fluctuations. Pressure correlation contours often have integral scales that differ with flow direction. The integral scale is thereby defined as the distance over which, at the average, two pressure pulses become fully uncorrelated. In other words, it defines the maximum possible area over which a pulse may reasonably act. Often, these contours are complex and difficult to obtain because requiring a lot of measurements. Nevertheless, for slabs that are very large compared to the integral scales of the pressure pulses, the spatial distribution of the time-averaged surface pressure field constitutes a plausible alternative to pressure correlation contours (Melo et al., 2006). Hence, no detailed regarding the contours is needed. Figure 1 compares the time-averaged mean dynamic pressures with the instantaneous total dynamic pressures. For large slabs, the large number of pressure peaks and pressure spikes compensate each other and spatial integration of instantaneous total pressures corresponds quite well to the blue surface, i.e. the time-averaged dynamic pressure field.

t 0 C total (x 0,t 0 ) = C p (x 0 ) + C p (x 0,t 0 ) x 2 x 2 C p (x,t 0 ) dx = C p (x,t 1 ) dx ~ 0 x 1 x 1 C p (x) integral scale x2 t 1 C p (x) dx x1 x1 (C p (x 1 ) + C p (x 2 )) ( 2 ) x ( (C (x, t ) + C (x, t )) tot 1 0 tot 2 0 ) x 2 Transient waves x2 Figure 1. Instantaneous versus Time-averaged Dynamic Pressures along Upper Face of Concrete Slab Second, slab uplift may be generated by pressures building up underneath the slab, in the confined area between the slab and its foundation. Transfer of pressures through the joints in between and underneath the slabs may then be considered in three ways: 1. Time-averaged dynamic pressure field: the pressure field underneath the slab is solely defined by the time-averaged values of dynamic pressures acting at the entrance of the joints between the slabs (Melo et al., 2006). This is illustrated by the blue lines in Figure 1. 2. Instantaneous dynamic pressure field: the pressure field underneath the slab is defined by both the time-averaged and the fluctuating part of the dynamic pressures acting at the entrance of the joints between the slabs (Fiorotto and Rinaldo, 1992). This is illustrated by the red lines in Figure 1. 3. Transient dynamic pressure field: the pressure field underneath the slab is defined by the time-averaged and fluctuating part of the dynamic pressures at the entrance of the joints between the slabs and by transient waves propagating through the joints (Bollaert, 2004). This is illustrated by the orange line in Figure 1. Whether or not pressure waves may have an influence largely depends on the assumptions made on the wave celerity. Two main approaches exist:

1. At high wave celerities, i.e. O(10 2-10 3 ) m/s, pressures travel quasi instantaneously through the joints and a net uplift pressure on the slab is the result of the difference between an instantaneous surface pressure field and its corresponding instantaneous underpressure field. Transient effects are neglected because occurring (and damped out) too fast for the turbulence at the surface. As such, the underpressure corresponds to the red line in Figure 1 and solely defined by instantaneous pressure pulses at the joint entrances. This is called a dynamic approach (Fiorotto and Rinaldo, 1992). 2. At low wave celerities, i.e. O(10 1-10 2 ) m/s, and for large slab lengths of O(10 1 ) m, a pressure wave needs time to be transferred all under the slab. When considering the joint as a resonator volume and the pressure pulses at the joint entrances as exciters, transient oscillations and even resonance conditions may occur, depending on the fundamental resonance frequency of the joint (Figure 2). For joint resonance frequencies that are close to the main frequencies of the impacting turbulent flow, pressures may amplify. Hence, underpressures are not only determined by instantaneous pulses at the joint entrances but also by the transient characteristics of the joints. This is called a transient approach (Bollaert, 2004). Figure 2. Main Resonating Frequencies of Lining Joint Excited by Surface Pressures Turbulent flow in stilling basins mainly occurs at rather low frequencies, i.e. a few Hz to tens of Hz (Toso and Bowers, 1988). To generate transient pressures, wave celerities have to be low and joint lengths have to be significant. Recent research (Bollaert, 2003) has shown that waves may travel at celerities that are very low, i.e. 50-200 m/s. This is due to the presence of free air in the flow mixture and is directly responsible for the appearance of transient effects in joints. Small-scale joint lengths and no air presence are the main reason that laboratory-scaled experiments are not able to generate such effects. The transient approach needs a quantification of pressure amplification inside the joints. This may be done in two ways: 1) by use of an appropriate pressure amplification coefficient (Bollaert, 2004); or 2) by direct numerical simulation of transient two-phase underpressures as a function of a time-dependent surface pressure field. The latter may be measured in the laboratory. Finally, a so generated net uplift pressure (force) may move the slab. For the most common case of anchored slabs, both the slab weight and the anchor stresses will prevent the slab from moving. This results in a dynamic equilibrium that is very similar to a spring-mass system as expressed by Newton s law (Fiorotto and Salandin, 2000). For such a system, the persistence time is of importance.

Differential equation for dynamic slab movement Based on Fiorotto and Salandin (2000), dynamic uplift of anchored concrete slabs may be expressed by the differential equation valid for a spring-mass system with a forced vibration by means of an external forcing function. Damping effects are neglected. This is a safe-side assumption that has its merit when using dynamic pressures. For transient pressures, however, damping effects have to be accounted for because highly fluctuating as a function of the amount of air inside the joints (Bollaert, 2003). The basic equation expresses a balance of stabilizing and destabilizing forces as a function of time: F stabt Fdestabt (1) Stabilizing forces are the slab mass m and the stresses induced in the steel anchors, based on their dynamic stiffness k. The equation then becomes: m z ' ' t k zt F t (2) The slab mass is defined by the concrete density c and the height of the slab h s. The dynamic stiffness of the steel anchors is determined by the steel elastic modulus, the steel sectional area A st and the length of the anchors L st. The equilibrium may be written per unit of slab surface as follows (valid for positive displacements z(t)): E destab '' st st h z t zt pt (3) c s in which p(t) stands for the net uplift pressure on the slab. Solving this equation as a function of time expresses the uplift of the slab governed by the inertia of its mass and the stiffness of its anchors. During slab uplift, the uplift pressure pulse is assumed constant. Also, the elasticity of the water and the underlying rock are neglected. The standard solution of this 2 nd order linear differential equation with constant coefficients may be written: L A st C2 C2 zt cos C1 t (4) C1 C1 The solution consists of the sum of two periodic motions at a different frequency but for the same amplitude. At present, the first motion on the right hand side of Equation (4) is constant (zero frequency) and the second motion on the right hand side is cosinusoidal with a frequency and a period T that equal: k C 1 (5) m

m 2 T 2 (6) k C 1 In other words, the function z(t) reaches its maximum value at time: t (7) C 1 This corresponds to a maximum displacement of the slab (per unit of area) of: The corresponding maximum possible steel stress is written: z C p t L 2 st max 2 2 (8) C1 Est Ast zmax pt E 2 (9) st, max st 2 Lst Ast st, static As pointed out by Fiorotto and Salandin (2000), dynamic equilibrium thus results in steel stresses st,max that are twice as high as the static steel stress st,static. However, this dynamic steel stress can only be reached provided that the persistence time of the net uplift pressure equals or exceeds the time period T needed to build up the cosinusoidal motion of the slab. Also, the dynamic steel stress becomes equal to the static steel stress already at time instant T/2. Hence, the persistence time of the net uplift pressure is essential. Short-duration pulses are not able to develop the full cosinusoidal motion and, thus, static steel stresses are valid. Pulses of longer duration allow the cyclic motion and dynamic stresses to fully develop. NEW DESIGN METHOD FOR CONCRETE LINING UPLIFT Types of flow impact The new design method for uplift of concrete linings presented hereafter has been developed for two types of flow impact commonly encountered in plunge pool stilling basins: 1) hydraulic jump turbulent flow, and 2) falling jet turbulent flow. For falling jets, the flow conditions in the stilling basin are determined by the flow conditions at issuance of the jet and modified during the fall of the jet. Dam issuance conditions are defined by the outlet structure, the upstream head and energy losses. The principal forces that act on a jet during its fall are gravitational contraction, spread due to turbulence and air drag (Ervine and Falvey, 1987). These allow determination of the

exact point of impact of the jet, as well as the geometric and hydraulic characteristics of the jet at this location. For hydraulic jumps, the flow conditions have to be determined at the beginning of the jump, i.e. near the toe of the dam. The flow conditions are first computed along the downstream face of the dam crest down to the dam toe. This defines the inflow conditions for the hydraulic jump. The main parameters of interest are the average flow velocity, the flow height and the Froude number at start of the jump. Dynamic pressures over the lining surface Falling jets and hydraulic jumps develop dynamic pressures that continuously fluctuate over the upper face of the lining of the stilling basin. As discussed before, the instantaneous pressures acting over the slab upper face can be approximated by their time-averaged values provided the slabs are large compared to the integral scales of the pressure fluctuations. Integral scales can be derived from available near-prototype scaled laboratory tests of high-velocity jet impact on slab joints (Bollaert and Schleiss, 2005). When a physical model is available, however, they may also be defined based on laboratory pressure fluctuation measurements. Integral scales can then be estimated by a correlation function based on data from the physical model, but n approached by an exponential law n e, in which n stands for the characteristic length and is a calibration coefficient (Fiorotto and Rinaldo, 1992). For hydraulic jumps, the characteristic length is equal to the incoming flow height. For jets, the characteristic length is equal to the jet diameter at impact. For jets, integral scales may be considered independent of flow direction, while for hydraulic jumps, the transversal integral scale is 5 to 6 times larger than the longitudinal one (Fiorotto and Rinaldo, 1992). The numerical grid used to compute the dynamic surface pressures over the slabs should not be coarser than the smallest integral scale of the pressure fluctuations. An example of a grid is presented in Figure 3 for a stilling basin in the US. The slabs are numbered alphabetically. A detailed grid is shown for slab W. Integral scales were in the order of 1.5 m, i.e. one order of a magnitude smaller than the slab sizes. The parameters of interest are the mean dynamic pressures and the root-mean-square (RMS) and extreme values of the fluctuating dynamic pressures, as well as their 2D spatial extension. These values can be assessed by means of pressure coefficients. These coefficients are obtained by dividing the absolute pressure values (in [m]) by the incoming kinetic energy of the flow (V 2 /2g, in [m]). For falling jets, pressure coefficients can be estimated based on available laboratory experiments and corresponding theoretical developments (Ervine et al., 1997; Melo, 2002; Bollaert, 2002; Melo et al., 2006). Similarly, for hydraulic jumps, dynamic pressures can be assessed based on literature (Toso and Bowers, 1988; Fiorotto and Rinaldo, 1992). Theory and mathematical expressions can be found in detail in Bollaert (2005).

Y-spacing 1.5 m surfaces 2.32 m 2 1.16 m 2 0.58 m 2 1.5 m X-spacing [m] 73.8 U V W X AP AQ AR 58.5 52.1 Q R S T AM AN AO M N O P AJ AK AL 36.9 Flow direction Y I J K L AG AH AI 21.6 E F G H AD AE AF 6.4 Z A B C D AA AB AC 0 x 372.2 387.4 X 402.6 417.9 433.1 448.4 463.6 478.8 [m] Figure 3. Numerical Grid Used to Compute Dynamic Pressures on a Stilling Basin Lining of a US Dam. Moreover, for hydraulic jumps, the pressure field is generally considered homogeneous in the lateral direction. For jets, however, the pressure field is considered twodimensional. Figure 4 presents the theoretically defined RMS dynamic pressure field for 3,200 m 3 /s jet impact flow conditions at a dam in the US. Dynamic underpressures between the lining and its foundation Computational method: The underpressure field may be computed by determining the dynamic surface pressures that act on the joints between the slabs or on fissures created in the concrete of the slabs and by supposing that these pressure travel through the joints/fissures (by failure of the water stops). A safe side assumption is to consider the maximum possible pressures that may act along the surface along the joints. A more realistic assumption is probably to consider the mean dynamic surface pressures. The methodology proposed in here to define net uplift pressures and impulsions on a slab is of pseudo-2d character because performed separately along the X and Y directions in a 1D manner (corresponding to the orthogonal joint directions in a Cartesian coordinate system). Potentially positive (pressure releasing) influences of the drainage system between the slabs and the foundation are safely neglected.

5 6 1 7 2 8 9 10 3 4 Elevation (ft a.s.l.) 500 450 H = 418 ft Q = 115 000 cfs 400 350 300 250 260 [ft a.s.l.] 240 220 200 200 180 160 150 140 120 100 100 900 950 1000 1050 1100 1150 1200 1221 1250 1271 1300 1321 1350 1371 1400 1421 [ft] 1450 1500 1550 1600 a) Distance (ft) b) Figure 4. RMS Pressure Fluctuation Coefficients Generated by Multiple Jets Impacting the Stilling Basin Lining of a US dam: a) Plan view; b) Side View and Perspective View. The maximum dynamic pressure coefficients at the slab surface (C max ) are first spatially averaged along each of the two opposite slab joints in both the X and the Y-direction (Figure 5). These C max,average coefficients account for lateral diffusion of local pressure peaks through the slab joints. Then, the underpressure field is formed by taking the mean value of the two C max,average values and by applying an amplification factor that accounts for transient effects. Next, this corrected value is applied to a percentage of the total area underneath the slab (Figure 5). Due to 2D diffusional effects of pressure waves, application to the total area would be far too conservative. The considered area has a length that is equal to the joint length L j in the perpendicular direction and a width W j as defined by a 2D calibration that is explained in detail hereafter. The process is performed in both the X and Y directions separately; the most critical result is retained. Determination of the width W j of the 1D strip that results in the exact total force under the slab can be done if measurements of net uplift forces on similar slabs are available. For spillway flow, such direct force measurements on 2D slabs are available from physical model tests (Bellin and Fiorotto, 1995). When subtracting the time-averaged spatially distributed dynamic surface pressure field from these the strip width can be defined. The 1D approach is thus calibrated based on 2D model tests for hydraulic jumps. It is assumed that this relationship holds for all possible flow conditions in the basin. Bellin and Fiorotto (1995) describe the net uplift force on a slab as follows:

F max I y l x x y t,, C p Cp lx ly 1 x l y 2 V 2g (10) in which is function of the instantaneous spatial pressure distribution over the total slab surface, C + p and C - p are the positive and negative dynamic pressure coefficients, is the specific weight of the water, and finally l x and l y are the slab dimensions in the X-and Y- directions respectively. depends on the shape of the slabs and on the ratios of the slab length to the integral scales x, y in both X and Y directions. Bellin and Fiorotto (1995) provided direct experimental evaluation of the uplift coefficient for a wide range of slab shapes and Froude numbers of the incoming flow field. This was performed by simultaneous measurements of pressures underneath simulated slabs and net uplift forces on the slabs. For jet impact, however, no such measurements are available in literature. It is proposed to use the values obtained for hydraulic jumps with similar Froude numbers and ratios of slab length to integral scales. The values highly depend on the ratios l y / y and l x / x. For very small and very high ratios, theoretically tends towards zero. In between, a maximum value is obtained for ratios between 2 and 4, assuming that maximum pulses occur at both slab joints and a minimum pressure occurs in between. The ratios tested by Bellin and Fiorotto (1995) equal 0 to 2 for l y / y and 0 to 10 for l x / x. C max,1 L j aver C max,1 aver C max,2 aver C max,1 C max,2 aver C max,2 W j L j X aver C max,2 aver C max,2 E st, A st, L st L c,h c, c aver C max,1 Y aver C max,1 Figure 5. Determination of Pressures Acting Underneath the Slab Based on the Average Values of Maximum Pressures Along Opposite Surface Joints Transient excitation frequencies and related steel stresses: The fundamental resonance frequency of a joint f res is a function of the wave celerity c and the joint length L j (Figure 2). The inverse T res expresses the the average persistence time T res of a pressure pulse. For example, for celerities of 100-500 m/s and joint lengths of 10-20 m, T res is written:

T 1 f 2 L c j res 0.04 0. 4 res s (11) For steel anchors that are 2 m long and slabs that are 1.5 m high and 15 m long, and for a steel area density of 4 cm 2 /m 2, the time periods necessary for pressure pulses to reach the static and dynamic steel stresses are 0.015 sec respectively 0.03 sec. In this case, pressure waves through the joints have a persistence time that easily allows reaching the dynamic steel stresses in the anchors. If this is not the case, static stresses might be more realistic. Probability of occurrence: The probability of occurrence of extreme pressure pulses is defined based on a Gaussian probability distribution for low and intermediate pulses and a Type-I probability distribution for extreme positive pulses (Toso and Bowers, 1988). Expressing Z as (X-)/, in which X are the pressure values, stands for the mean pressure value and for the standard deviation of the pressure fluctuations, the following probability density distributions can be defined: p Z 1 e 2 2 Z 2 for Gaussian distribution (12) p Z e Z Z e e for Type I distribution (13) Based on physical model experiments, it has been found that, for positive Z-values of up to about 2, a Gaussian distribution may be used. For higher Z-values, a Type I distribution seems more appropriate. A comparison with probabilities derived from pressure records measured on the physical model shows that the Type I distribution may be used up to Z values of around 12-13. At higher Z values, the pressure pulse is assumed too localized in space and of no direct relevance to slab uplift. Numerical computations of transient pressures under a lining: A 1D transient two-phase numerical model has been developed to compute underpressures between the slabs and the underlying foundation. Detailed characteristics of the numerical model and the used equations can be found in Bollaert and Schleiss (2005). The model needs pressures measured at the joint entrances, for example on a physical model, as input data. This is presented in Figure 6. The model defines the force under the slab as: F p A (14) u i ui ui in which A ui stands for the area of application of p ui. For sake of simplicity, each area of the five points p ui has been taken equal to 1/5 th of the total slab area. The uplift force is then computed as the arithmetic average of the underpressures times the total slab area. The influence of transient pressure waves on the net uplift forces on the slabs can be expressed by means of a transient amplification factor defined as the ratio of the average underpressure to the average of the maximum surface pressures p sj :

Pressure [m] i i (15) p j j The amplification factor only accounts for the underpressures and not for the dynamic surface pressure field. It defines the maximum amplification that the underpressures may exhibit due to transient wave effects in the joint. Hence, for a transient slab uplift computation, the dynamic underpressures are simply multiplied by this amplification factor to obtain fully transient values (Figure 5). p ui sj p s1 p s2 p u1 p u2 p u3 p u4 p u5 p s1-2 = average value of maximum surface pressures along the joints p u1-5 = computed transient pressure values underneath slab Figure 6. Methodology of Numerical Computation of Slab Uplift Pressures. 200 150 100 Input pressure ps1 Input pressure ps2 Average of computed underpressures psu Average transient underpressure psu Surface peak ps1 Surface peak ps2 50 0 0 0.1 0.2 0.3 0.4 0.5 0.6 0.7 0.8 0.9 1 Time [sec] Figure 7. Result of Transient Pressure Computations For a Jet Impacting a Lining. Figure 7 shows the result of a numerical computation of transient pressures acting underneath a lining. The pressures acting at the joint entrances have been deduced from physical model experiments of a dam in the US. When pressure pulses enter quasi simultaneously at both joint entrances (surface peaks 1 and 2), an amplification of these pressures is obtained over the whole joint length (transient peak). Transient waves are amplified only at low frequencies, i.e. maximum a few Hz. Also, all computed pressure

pulses have a duration of minimum 0.10 sec, much higher than the minimum time period of 0.03 sec. required to double the steel stresses. Second, higher frequencies are of no influence to the transients. This is in agreement with the fundamental resonance frequency of the assumed open-open boundary resonator system. For a joint length of 18 m and an average wave celerity c of 100 m/s, one obtains a resonance frequency < 3 Hz. The amplification factor has been computed for two different celerity-pressure relationships, corresponding to low (0-2 %) and high (5-10 %) air concentrations in the slab joints and directly defining the damping of the transients (Bollaert, 2003). For practice, amplification values of 1.35 for jet impact and 1.20 for hydraulic jumps have to be used. Computational Methodology 1. Determine integral scale of pressure fluctuations (based on theory and/or physical model experiments) and check plausibility of time-averaged surface pressures. 2. Determine pressure coefficients along upper face of lining (numerical grid) 3. Choose initial slab dimensions and joint/fissure locations 4. Determine pressure coefficients along joints between the slabs of the lining or along fissures through the lining 5. Determine joint/fissure length -averaged values of the maximum dynamic surface pressures for the X (longitudinal) and the Y (transversal) direction. 6. Determine width of the 1D strip based on Bellin and Fiorotto (1995). 7. Multiply the so defined underpressures/forces by a transient amplification factor to account for transient wave effects. 8. Determine fundamental resonance frequency and average persistence time of pressure pulses. Compare with time duration necessary for dynamic steel stresses to develop. 9. Choose final slab dimensions and determine necessary steel area based on the spring-mass equation and allowable elastic steel stresses. Based on the above design method, Figure 8 shows the steel stresses that may develop in the anchor bars of a new plunge pool lining design for a high-head arch dam in Switzerland. The computed steel stresses are presented as a function of the longitudinal location of a transversal fissure that has been imagined in the lining and that allows transfer of jet impact pressures underneath the lining. The blue surface corresponds to the total slab area between two construction joints. CONCLUSIONS AND RECOMMANDATIONS The present paper illustrates a new method for designing concrete linings of stilling basins against sudden uplift by impact of turbulent flows. The method is valid for any type of turbulent flow environment provided that the turbulent pressures of that flow can be statistically described by means of pressure coefficients.

Steel stresses [MPa] Construction joint Construction joint Zone of fissure locations 0-200 Zone -400-600 Jet impact area -800-1000 -1200-1400 -1600-1800 -2000 Anchor bars D = 20 mm; e = 1.5 m Anchor bars D = 36 mm; e = 1.0 m Anchor bars D = 40 mm; e = 0.5 m Construction joints Admissible steel stress 0 1 2 3 4 5 6 7 8 9 10 11 12 13 14 15 16 17 18 19 20 21 22 23 24 25 26 27 28 Distance from dam [m] Figure 8. Steel Stresses Computed as a Function of the Position of a Fissure in the Lining and For Different Anchor Bar Diameters and Spacings. The method uses time-averaged dynamic pressures along the upper face of the lining and supposes a transfer of peak pressure pulses through the joints to determine the pressures acting along the lower face of the lining. Based on detailed laboratory measurements of net uplift forces on slabs for different flow conditions (Bellin and Fiorotto, 1995), the total underpressure can be derived by adding the time-averaged surface pressure field to the laboratory measurements of the net uplift pressure. As such, the peak pressure values that are supposed to act under the slabs are applied over a restricted area of the slab to comply with the total underpressure measured in the laboratory. Because the small-scale laboratory tests did not account for transient waves, these peak pressure pulses have then to be multiplied by an amplification factor accounting for transient wave effects through the joints. Finally, subtracting the computed surface pressure field from the corrected underpressures results in transient net uplift pressures and forces on the slabs. Use of a differential equation valid for a spring-mass system then allows dimensioning the necessary steel area of the slab anchors. The method has already been applied on real-life studies of stilling basin design in both the western US and central Switzerland. The new lining of the latter stilling basin is actually under construction. REFERENCES Bellin, A.; Fiorotto, V. (1995): Direct dynamic force measurement on slabs in spillway stilling basins, J. of Hydr. Engng., ASCE, Vol. 121, N HY 10, pp. 686-693. Bollaert, E.F.R. (2003): The influence of joint aeration on dynamic uplift of concrete slabs of plunge pool linings, XXXth IAHR Congress, Thessaloniki, Greece, 2003.

Bollaert, E.F.R. (2004): A new procedure to evaluate dynamic uplift of concrete linings or rock blocks in plunge pools, Proc. of the Intl. Conf. Hydraulics of Dams and River Structures, Yazdandoost & Attari (eds.), Teheran, Iran, pp. 125-132. Bollaert, E.F.R.; Schleiss, A. (2005): Physically Based Model for Evaluation of Rock Scour due to High-Velocity Jet Impact, J. of Hydr. Eng., Vol. 131, N 3, pp. 153-165. Bowers, C.E.; Tsai, F.Y. (1969): Fluctuating pressures in spillway stilling basins, Journal of Hydraulics Division, ASCE, Vol. 95, N 6, pp. 2071-2079. Ervine, D.A.; Falvey, H.R. (1987): Behavior of turbulent jets in the atmosphere and in plunge pools, Proceedings of the Instit. of Civil Engineers, Part 2, Vol. 83, pp. 295-314. Ervine, D.A.; Falvey, H.R.; Withers, W. (1997): Pressure fluctuations on plunge pool floors, Journal of Hydraulic Research, IAHR, Vol. 35, N 2. Fiorotto, V.; Rinaldo, A. (1992): Fluctuating uplift and lining design in spillway stilling basins, Journal of Hydraulic Engineering, ASCE, Vol. 118, HY4. Fiorotto, V.; Salandin, P. (2000): Design of anchored slabs in spillway stilling basins, Journal of Hydraulic Engineering, ASCE, Vol. 126 (7), 2000, pp. 502-512. Liu, P.Q., Dong, J.R. and Yu, C. (1998): Experimental investigation of fluctuating uplift on rock blocks at the bottom of the scour pool downstream of Three-Gorges spillway, Journal of Hydraulic Research, IAHR, Vol. 36, N 1, 55-68. Lopardo, R.; Henning, R.E. (1985): Experimental advances on pressure fluctuations beneath hydraulic jumps, Proc. 21st Congress IAHR, Melbourne, Vol. 1, pp. 633-638. Melo, J.F. (2002): Reduction of plunge pool floor dynamic pressure due to jet air entrainment, Intl. Workshop on Rock Scour, EPFL Lausanne, Switzerland, pp. 125-136. Melo, J.F., Pinheiro, A.N., Ramos, C.M. (2006): Forces on plunge pool slabs: Influence of joints location and width, J. of Hydr. Engng., ASCE, Vol. 132, N 1, pp. 49-60. Reinius, E. (1986): Rock erosion, Water Power and Dam Construction 38(6), pp.43-48. Sanchez, B.J.S.; Viscaino, A.C. (1973): Turbulent effects on the lining of stilling basin, Proc. of the 11 th ICOLD Congress Madrid, Q. 41, Vol. 2. Toso, J.; Bowers, E.C. (1988): Extreme pressures in hydraulic jump stilling basin, Journal of Hydraulic Engineering, ASCE, Vol. 114, N HY8, pp. 829-843. Yuditski, G.A. (1963): Actual pressure on the channel bottom below ski-jump spillways, Izvestiya Vsesoyuznogo Nauchno Issledovatel Gidrotekhiki, Vol. 67, pp. 231-240.